Google’s Search Ranking Systems

Tuesday, April 23rd
Content Manager

Mark Wilson

Google rankings are the final boss - to use some video gaming terminology - of a lot of marketing efforts. While ranking highly on Google for relevant terms is far from the only step needed to ensure consistent traffic, leads and revenue, it’s a great start. Entire jobs are devoted to these SEO efforts.

To do that, it’s worth understanding how Google’s search algorithm works. The problem is that understanding it isn’t easy.

This isn’t simply complexity for its own sake. The complexity of Google’s search, and the many strategies to improve rankings, are a result of how varied search queries can be, and how those varied queries demand different types of responses from search engines.

So let’s look at Google’s ranking systems. And that’s right, I said systems, plural, which is the first step to understanding how search results work.

Google’s Search Rank Environment

Google’s search algorithm isn’t one system. This is the first myth we need to bust.

In fact, it’s numerous systems. We’ve listed the biggest subsystems and how they function below, but I don’t believe this article is a comprehensive listing of all ranking systems, particularly since Google is testing new models constantly.

For our purposes, though, it’s important to understand that search results are a product of dozens of subsystems working concurrently, and depending on the nature of the information being crawled and the search query, your website’s content may be affected by some of these systems more than others.

Retired Search Ranking Systems

Below are some major updates to the search algorithm that have reportedly been retired by Google.

Importantly, though, the insights and adjustments from each of them are often built into later iterations of ranking systems. So, for example, improvements from Hummingbird didn’t disappear, they were simply absorbed into later core system updates.


Hummingbird was released in 2013 and largely affected what has become known as Google’s Knowledge Graph. This allows the search engine to answer many questions without having to send the user to another website.

Hummingbird was also one of Google’s earlier and larger attempts to adjust for semantic meaning of search inputs. Stated differently, the search engine would try to interpret what you were searching for with a string of words. A query of “coffee” could pull up a website about the history of coffee, for example, but it’s more likely that the user is simply searching for nearby coffee shops.

This sort of interpretation is difficult for computers, but search engines have improved at it over time.


Released in 2011, Panda was a system designed to weed out low-quality content. It was largely in response to content farms that dominated SERPs at the time, generating massive quantities of low-quality content to brute-force their way into search rankings.

Many of Panda’s subsystems are now part of Google’s core algorithm.


Penguin released shortly after Panda, and was another adjustment intended to reward higher-quality content in SERP rankings, largely in response to keyword stuffing and spam that was populating many search results.

RankBrain Ranking System

RankBrain is a major Google search ranking system that was first released in 2015, and it’s still heavily in use into the 2020s.

RankBrain does several things, but its biggest innovation was in using AI learning to discover how individual words related to concepts, and how those concepts related to the intent of a search query.

If you type in “Arnold robot movie” you’re likely to get results for the Terminator films. But prior to this sort of search learning, it would have been hard for a search engine to produce the correct results, since what you were actually looking for wasn’t a term in your search input.

Further, what type of page are you looking for in regard to those movies? Do you want to stream them? See reviews? Or read a synopsis?

In this case, I don’t have a specific answer, but Google’s RankBrain system is able to learn over time to produce the most relevant results for the highest number of people. The processes involved in figuring this out for literally billions of searches are staggeringly complex, but that’s what RankBrain is able to do.

Information Systems

Not every Google listing is related to a specific and permanent webpage. Information related to topical or urgent issues can be delivered by the search engine, and there are subsystems designed to deliver accurate information when it’s needed.

Disaster Updates

Google has adjusted this system numerous times through the years, but the goal is to deliver real-time information on developing emergency situations.

This could be floods, fires, tornados, earthquakes, and more. Often, an information panel will appear at the top of the screen after a related search is run.

Personal Crises

This usually relates to medical emergencies that people are experiencing, and requires that they receive actionable information immediately.

This could include local emergency response contact information, or steps to take in a personal health emergency.

Passage and Snippet Ranking Systems

It might seem odd that Google also has subsystems for evaluating sections of pages rather than entire pages themselves. It does both, and there are benefits to passage-level analysis.

For instance, if you’ve ever read the “snippet” at the top of search results (usually a brief paragraph with a summary of the search query, often accompanied by a thumbnail), that’s a passage that was selected for its relevance.

Or if you’ve read one of the “People Also Ask…” questions as you scroll through search results, the response to those questions is a result of a passage analysis.


Multitask Unified Model, or MUM, isn’t used as part of general search results. Rather, it’s most often applied to snippet analysis that we mentioned just above, evaluating passages to highlight as the best short response to a search.

Passage Ranking

This will go into pages and assess passages individually. This helps Google understand the entire page’s relevance, and can also be pulled out for individual use.

Sometimes you’ll see a passage from an article as the text description of a search result, and this brief passage has replaced the official meta description that the site assigned to the page. This is an example of Google trying to determine the most relevant summary for a page, even if it’s from the body text of the page.

This is one of the reasons semantic HTML can be so important, to instruct search crawlers not just on a page’s intent but individual sections as well.

General Content Ranking Systems

With the exception of RankBrain, which is one of the larger core parts of Google’s current ranking system, we’ve danced around the bulk of Google’s analytical systems for the majority of searches.

Below are various ranking systems Google uses for varying purposes to deliver traditional search results.


BERT stands for Bidirectional Encoder Representations from Transformers.

Which is a fancy way of saying it learns what clusters of words mean. Hot dog, hot trends and hot day all have different meanings for “hot.” It’s BERT’s job to figure out what those are.


If you cover a particular industry, chances are you have a lot of pages devoted to similar topics. Oftentimes, you might even use identical web copy to describe something in two different places.

Google is dealing with duplicate content on a massive scale. Sometimes millions of pages will have the same information, particularly when it’s something codified like an historical speech.

Deduplication systems prioritize links with duplicate content, and also do things like ensuring the same website doesn’t show up twice in results with near-identical content from two different pages.

Exact Match Domain System

Imagine you start a coffee website and the URL is, and you have a subdomain of coffeenearme/coffee-near-me. Are you the best search result for someone in Boston looking for a local coffee shop? Probably not.

That’s more-or-less what exact match domain systems work on. URLs matter in terms of relevance of your content, but there are sites that will try to trick the system into listing them for searches where they aren’t the most useful result.

Freshness Systems

Freshness systems mostly come into play for topical subjects and news. If you have the internet’s best article on colon cancer research, it’s still going to be relevant in six months.

But if a restaurant just opened and you’re searching for reviews, you don’t want to read the article from two years ago announcing that it was under construction.

So the system both identifies subjects that should be prioritizing “fresh” content, and then focuses on the actual prioritization.

Helpful Content System

Shouldn’t all content be helpful in some way? Yes, ideally, but that’s not the world we live in.

There’s content that’s created to provide value for the visitor, and content that’s created solely to try to get search rankings, and doesn’t actually provide much value.

In practice, the processes for determining what “useful” means can get pretty complex, but these systems weed out content that is just spamming keywords or pages with the hope of achieving a good rank.


PageRank is an older system, but an important one and is still in use today. The core of its functionality has to do with link analysis.

Backlinks from other sites are a large indicator of trust. Similarly, the sites you link to often indicate your sources. And what you link to internally (i.e. within your own website) can teach Google what other content on your site is related to a topic.

Analyzing and ranking a site’s authoritativeness based on these factors is a large portion of what PageRank does, and it’s why quality, relevant backlinks are extremely important.

Site Diversity

Have you ever noticed that it’s difficult to rank for a search term more than once? This isn’t by accident.

With rare exceptions like business names, where a company’s homepage, About Us page, LinkedIn Page, Facebook page, and so on might all be ranked on Page 1, you’ll almost never see a site rank more than once for a term.

This is due to site diversity systems that prevent a single site from flooding results, giving people numerous quality options.

Spam and Demotion Systems

Everything we’ve discussed so far has mostly been geared toward ranking factors to help identify quality content.

The flip side of the coin is that Google needs to have systems in place that prevent scams, spam, and other low-quality content from ranking consistently.

As I think we all know to some degree, there’s a lot of junk on the internet. A lot. So how does Google deal with it?

They have a system, for instance, that monitors for content removals as a result of copyright claims, and is able to demote other content from that site.

Other systems will penalize sites that will post personal information, then charge a fee to have it removed (yes, these unfortunately exist).

It’s going to be hard to canvas every subsystem that’s designed to protect users and remove spam, because there are a lot of shady SEO practices that Google has had to contend with over the years.

In recent years, for instance, Google has had to crack down on content spam from AI tools that’s intended to do the same thing that keyword stuffing was doing more than a decade ago.

The point is that if you aren’t creating useful, reliable, accurate content, you’re probably a target for one or more of these removal or demotion systems.

Putting It All Together

Intimidated yet? It’s ok, it’s a lot to take in for anyone.

That said, the core tenets of creating good content are pretty straightforward, and will jive with pretty much any of the systems above if you’re executing your strategy thoroughly.

Keep the user in mind with the content you create. Have an intuitive, easy-to-use design and interface. Use SEO markers to indicate what a site, page, or passage is about.

There’s more we could discuss, of course (there’s always more), but that would take us beyond the point of this article: to give you a high-level overview of Google’s systems, how they interrelate, and how your website or app’s content interfaces with each system.

Understanding the system you operate within for SEO is the first step. The next is putting that knowledge into practice. If you’re ready to do that with a trusted guide, reach out to Leadflask today for a free, no-obligation consultation.

Additional Resources

Let's set up a call